Transcriptome-wide profiling and quantification of N6-methyladenosine by enzyme-assisted adenosine deamination

Abstract

N⁶-methyladenosine (m⁶A), the most abundant internal messenger RNA modification in higher eukaryotes, serves myriad roles in regulating cellular processes. Functional dissection of m⁶A is, however, hampered in part by the lack of high-resolution and quantitative detection methods. Here we present evolved TadA-assisted N⁶-methyladenosine sequencing (eTAM-seq), an enzyme-assisted sequencing technology that detects and quantifies m⁶A by global adenosine deamination. With eTAM-seq, we analyze the transcriptome-wide distribution of m⁶A in HeLa and mouse embryonic stem cells. The enzymatic deamination route employed by eTAM-seq preserves RNA integrity, facilitating m⁶A detection from limited input samples. In addition to transcriptome-wide m⁶A profiling, we demonstrate site-specific, deep-sequencing-free m⁶A quantification with as few as ten cells, an input demand orders of magnitude lower than existing quantitative profiling methods. We envision that eTAM-seq will enable researchers to not only survey the m⁶A landscape at unprecedented resolution, but also detect m⁶A at user-specified loci with a simple workflow.

Figure 1 |. Global A deamination by TadA8.20.

a. Proposed m⁶A detection scheme. TadA8.20 selectively converts A into I, without acting on m⁶A. I is recognized as G by reverse transcriptases. Persistent A post-TadA8.20 treatment corresponds to m⁶A. b, c. In vitro deamination of RNA probes hosting A or m⁶A in “CGAUC” (b) and “GGACU” (c) motifs by TadA8.20. Unmethylated and methylated RNA sequences were prepared through in vitro transcription using ATP and N⁶-methyl-ATP as starting materials, respectively. Treated RNA was reverse transcribed, amplified, and subjected to Sanger sequencing. d, e. TadA8.20-catalyzed A-to-I conversion rates in “CGAUC” (d) and “GGACU” (e) probes quantified by next-generation sequencing. f. Deamination of synthetic A/m⁶A RNA probes by TadA8.20. 53-nt RNA probes hosting NNANN and NNm⁶ANN motifs were treated by TadA-8.20. Deaminated RNA underwent RT and next-generation sequencing. g. Correlation of persistent A signals captured by eTAM-seq and m⁶A contents in RNA probes. h. Capillary gel electrophoresis analysis of fragmented HeLa mRNA treated with or without TadA8.20 at different temperatures for 3 h. RNA size distribution is plotted on the right. For eTAM-seq, RNA is incubated with TadA8.20 at 53°C for 1 h followed by 2 h treatment at 44°C. Experiments were repeated independently with similar results. i. Transcriptome-wide A-to-I conversion rates in two independent replicates. 10% of A sites with ≥100 counts were randomly sampled to make the scatter plot. Pearson’s r was calculated for all A sites with ≥100 counts. j. Two m⁶A sites in human rRNA. Positions 1829–1835 of 18S rRNA and positions 4217–4223 of 28S rRNA are plotted.

Figure 2 |. Transcriptome-wide m⁶A profiling in HeLa mRNA by eTAM-seq.

a. Overlap analysis of m⁶A sites identified in three biological replicates of eTAM-seq (HeLa/IVT). b. Methylation levels of common sites detected in eTAM-seq (HeLa/IVT-1) and eTAM-seq (HeLa/IVT-2). Correlative analyses on methylation levels reported by replicate 1 vs 3 and 2 vs 3 are provided in Supplementary Fig. 7. c. Hit distributions in DRACH and non-DRACH sequences at different methylation levels. d. Cumulative m⁶A signals from highly methylated sites to lowly methylated sites (right to left). m⁶A constitutes 0.41% of all A subject to evaluation in the HeLa transcriptome. e. Overlap analysis of m⁶A sites identified by eTAM-seq and peak clusters generated via MeRIP-seq. The overlap between eTAM-seq and MeRIP-seq increases with higher read depth and methylation levels. f. Metagene plot of transcriptome-wide distribution of m⁶A. m⁶A distributions across different RNA regions are provided in the inserted pie chart. IGR: intergenic region; ncRNA: non-coding RNA. g. Major sequence motifs hosting m⁶A. DRACH motifs are in black and non-DRACH motifs are colored in red. The consensus sequence hosting m⁶A is inserted. h. m⁶A sites co-discovered by eTAM-seq and m⁶A-SAC-seq. Hits in DGACU captured by both methods are subject to overlap analysis. m⁶A-SAC-seq dataset: GSE198246. i. m⁶A positions and fractions in MALAT1, TPT1, MYC, and ZBED5. eTAM-seq signals are plotted as methylation levels (%) alongside MeRIP-seq peaks in normalized read coverage. Note that eTAM-seq (HeLa/IVT) has slightly higher coverage than eTAM-seq (HeLa/FTO) and may therefore capture more m⁶A sites. MALAT1_2515, 2577, 2611 and TPT1_687, 703 are indicated by arrows. The coding sequence for TPT1 is on the minus strand of the genome.

Figure 3 |. m⁶A profiling in mouse embryonic stem cells (mESCs) by eTAM-seq.

a. Hit distributions in DRACH and non-DRACH sequences across different methylation levels. m⁶A sites identified by eTAM-seq (mESC/IVT) are plotted. b. Metagene plot of transcriptome-wide distribution of m⁶A. m⁶A distributions across different RNA regions are inserted. IGR: intergenic region; ncRNA: non-coding RNA. c. Overlap analysis of m⁶A sites identified by eTAM-seq and peak clusters generated via MeRIP-seq. Hits detected by eTAM-seq (mESC/IVT) are overlapped with a published MeRIP-seq dataset (left). One MeRIP-seq peak covers multiple m⁶A sites (right). Similar analyses using the eTAM-seq (mESC/FTO) dataset are provided in Supplementary Fig. 18. d. Methylation levels reported by eTAM-seq (mESC/IVT) and eTAM-seq (mESC/FTO). e. m⁶A positions and fractions in selected regions of Nanog, Sox2, and Klf4. eTAM-seq hits are plotted in methylation levels (%) and are juxtaposed with MeRIP-seq peaks in normalized read coverage. For a zoomed-out view of m⁶A distribution in full-length Nanog, Sox2, and Klf4, see Supplementary Fig. 19.

Figure 4 |. m⁶A is strongly depleted in Mettl3 KO mESCs.

a. Venn diagram showing the overlap of eTAM-seq-detected m⁶A sites in ctrl and Mettl3 KO mESCs. b. Hit distributions in DRACH and non-DRACH sequences. c. Methylation levels of eTAM-seq-captured m⁶A sites in ctrl and Mettl3 KO mESCs. Lower and upper hinges in the box plot represent first and third quartiles with the center line and red dot representing the median and the mean, respectively. Whiskers cover ±1.5× of the interquartile range. d. Scatter plot of methylation levels for m⁶A sites jointly identified in ctrl and Mettl3 KO mESCs. e. Changes of methylation levels in ctrl and Mettl3 KO mESCs. Methylation difference for a given A site = methylation level in ctrl mESCs – methylation level in Mettl3 KO mESCs.

Figure 5 |. m⁶A impacts transcript stability.

Cumulative distributions for transcripts of different half-lives in HeLa cells treated with control and METTL3-targeting siRNA. Transcripts methylated to different levels are analyzed in separate bins (high m⁶A: n = 1,593; medium m⁶A: n = 1,594; low m⁶A: n = 1,593; no m⁶A: n = 1,758). Box violin plots of transcript half-lives are inserted. Lower and upper hinges represent first and third quartiles. The center line and the red dot denote the median and the mean, with whiskers covering ± 1.5× of the interquartile range. P-values were determined by one-tailed Wilcoxon rank-sum test using the unmethylated group as a reference. HeLa mRNA half-life dataset: GSE49339.

Figure 6 |. Site-specific, deep sequencing-free m⁶A detection and quantification.

a. Workflow for eTAM-seq-enabled site-specific quantification of m⁶A. mRNA is fragmented, ligated to a DNA adapter, treated by TadA8.20, and reverse transcribed into cDNA. Site-specific primers are designed to recognize post-deamination RNA sequences and amplify the loci of interest. m⁶A quantification can be achieved by both Sanger sequencing and amplicon deep sequencing. b. Quantification of methylation levels for 8 m⁶A sites in HeLa mRNA by Sanger sequencing, amplicon deep sequencing, and RNA-seq. Tad8.20-treated IVT samples are provided for reference only. c. Methylation quantification for ACTB_1427 and EIF2A_994 with 5 ng, 500 pg, 50 pg, and 5 pg mRNA. d. Methylation quantification for ACTB_1427 and EIF2A_994 with 25 ng, 2.5 ng, and 250 pg total RNA.